marc mayol
Projects
Talks
Blog
About
es
Contact
Learn & get
informed.
Articles where I share what I learn about generative AI, Google tools and hands-on AI development.
Chunking in RAG: strategies, overlap, chunk size and how to evaluate it
Chunking strategies for RAG: fixed-size, recursive, sentence, semantic, structure, late and agentic. How to pick chunk size and overlap and evaluate it.
28 Oct 2026
Gemma on Android: running an on-device LLM with LiteRT
How I got Gemma running on an Android phone: exporting the model to .litertlm, wiring it into the app and what performance to expect, mistakes included.
28 Oct 2026
LLM observability: comparing Langfuse, Opik, LangSmith, Phoenix and more
A neutral comparison of LLM observability tools: Langfuse, Opik, LangSmith, Phoenix, Helicone and MLflow. Licence, pricing, tracing and evaluation.
28 Oct 2026
Run an LLM locally: a practical guide to on-device AI
When running an LLM locally pays off, what hardware you need, how to pick the model and quantization, and the real limits I hit on PC and on Android.
28 Oct 2026
Exercise API: an open, free, bilingual database of gym exercises
How Exercise API was born: a free, open API with 100+ gym exercises, illustrations and muscle maps, served as static files from GitHub Pages.
16 Jul 2026
JSConf Spain 2026: I Had a Blast (And Learned a Ton)
My take on JSConf Spain 2026: talks by Javi Velasco and Jaime Gómez-Obregón, and Alan Buscaglia's workshop. An event I'd recommend to any developer.
17 Mar 2026
Prompt-First: Complete Guide to the New AI Development Paradigm
A guide to prompt-first development: building software by describing what you want in prompts instead of writing the code, with examples.
16 Nov 2025
Shadow AI: The Hidden Artificial Intelligence Lurking in Your Company
Shadow AI is staff using tools like ChatGPT without company approval. The risks (data leaks, GDPR fines), the Samsung case and how to manage it.
12 Nov 2025
Langfuse: The Definitive LLM Observability Platform in 2025
Langfuse, an open-source LLM observability platform: full tracing, automatic metrics and self-hosting, with LangGraph and LangChain integration.
5 Nov 2025
Opik: Your Copilot for Mastering LLMs in Production
Opik is an open-source LLMOps tool to trace LLM apps in production, evaluate prompts, catch hallucinations and keep costs under control.
4 Nov 2025
MCP vs Claude Skills
MCP or Claude Skills: what each one does, a comparison table, a sample SKILL.md and MCP server config, and when to use one, the other or both.
2 Nov 2025
Anthropic Agent Skills. What they are, how they work and why they matter for enterprise AI
What Anthropic's Agent Skills for Claude are, how they load instructions and code only when needed, and how to use them for real business tasks.
28 Oct 2025
Stanford's trick to make AI learn on its own
How Stanford's ACE (Agentic Context Engineering) lets AI agents improve by reflecting on and updating their own context, with no retraining.
21 Oct 2025
Chonkie and Docling: two complementary approaches to document processing in RAG pipelines
Chonkie or Docling for RAG pipelines: what each one does, how they differ and how to combine them to process your documents better.
17 Oct 2025
Chonkie: how to chunk text for RAG with chunkers and Pipeline
A guide to Chonkie, the MIT library for chunking text in RAG: chunkers, Pipeline, refineries and vector databases, with code tested on version 1.7.
16 Oct 2025
Hexagonal architecture and language models: flexibility and robustness in AI
How hexagonal architecture (ports and adapters) helps you build LLM applications that can switch models or providers without rewriting the core.
15 Oct 2025
Event-Driven Architecture and LLMs: How Effective Is This Combination?
Event-driven architecture and LLMs: advantages, challenges, and key points to apply EDA in artificial intelligence with scalability and resilience.
3 Oct 2025
Docling: the easy way to work with files and LLMs
Docling, IBM's open-source tool, turns PDFs, Word files and images into clean text for LLMs. What it does and how to start using it.
14 Sept 2025
Google reshapes the AI landscape at Cloud Next 2025
A roundup of the AI announcements Google made at Cloud Next 2025 and what they mean for teams building on Google Cloud.
11 Apr 2025
Firebase Genkit
At Jsconf 2025, held in Madrid, I discovered an interesting library during one of the workshops: the Firebase Genkit framework.
18 Mar 2025
Gemini Robotics: Google’s leap into intelligent robotics
Google, through its DeepMind lab, continues to drive the future of artificial intelligence and robotics with Gemini Robotics and Gemini Robotics-ER
18 Mar 2025
Google Transforms Image Generation with Gemini 2.0 Flash
Google Gemini 2.0 Flash transforms image editing with native AI. Discover its features and how to try it for free.
18 Mar 2025
Model Merging in AI: Potential, Tools and Considerations
What model merging is in AI, which tools such as mergekit are used, and what to keep in mind before combining language models.
11 Oct 2024